t5-base
by google-t5
223M params · translation · 778 likes · 2.6M downloads
t5-base is a 223M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (5890 tok/s), GeForce RTX 5090 (8833 tok/s), M4 Max 128GB (2155 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 5890 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 8833 tok/s | comfortable |
| M4 Max 128GB | 128GB | 2155 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 1077 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 1077 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 10792 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 21706 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 5208 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 4306 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 4180 tok/s | comfortable |